Theoretical analysis of perturbation multi-dividing ontology learning algorithm

The multi-dividing ontology learning algorithm is specially designed for tree-structured ontology graphs, and has become a paradigm of graph-based ontology learning. In view of the disturbance of ontology data, this paper proposes perturbation multi-dividing ontology learning approach. Assuming that the perturbed ontology data are drawn from the same distribution as before, the error bound of perturbation multi-dividing ontology learning is given in such hypothesis. Finally, we analyze flaws in theoretical results and gaps with practical applications, and raise the open problem for future study.